GodeX: OpenAI Responses API gateway for GLM/Zhipu Chat Completions

[Ahoo-Wang]

Hi GLM community,

I wanted to share a small open-source integration for developers who use Codex-compatible clients but want to route requests to GLM/Zhipu Chat Completions models through a local gateway.

GodeX exposes an OpenAI Responses API compatible endpoint locally and bridges it to Chat Completions providers, including Zhipu/GLM. The project focuses on protocol compatibility rather than replacing provider SDKs:

• request normalization from Responses input items into chat messages
• tool declaration planning and tool call restoration
• streaming event reconstruction back into Responses-style SSE events
• previous_response_id session persistence on the gateway side
• compatibility diagnostics when a Responses feature needs to be degraded, ignored, or rejected
• request tracing for debugging provider/model behavior

Project: https://github.com/Ahoo-Wang/GodeX
Docs: https://godex.ahoo.me/
Chinese docs: https://godex.ahoo.me/zh/

This is not a bug report; it is an integration note for developers testing GLM models in Codex-style workflows. Feedback on GLM-specific compatibility behavior would be welcome.